[RIVER-110] Calling destroy should stop in progress blocked queries

• Type: Improvement
• Status: Open
• Priority: Minor
• Resolution: Unresolved
• Component/s: com_sun_jini_outrigger
• Bugtraq ID: 6249489
• Assignee: Unassigned
• Reporter: Frank Barnaby
• Created: 27/Jul/07 15:03
• Updated: 27/Jul/07 15:03

Description

Bugtraq ID 6249489

By default Outrigger will wait for up to 2 minutes for in progress remote calls to terminate before forcing an unexport of the server. This is done so the destroy call is given a chance to return - however with the 2.0 query protocol blocked queries will also keep the unexport(false) from succeeding. Since each blocked query is represented by a Watcher it should be possible (by adding a call new method to QueryWatcher) to stop each query and force it to return with an exception.

Comments
If we want Outrigger's VM to exit cleanly and quickly after DestroyThread.run() is done when there are in-progress blocked queries we need to address this.

Work Around
Setting maxServerQueryTimeout to a small value provides a partial workaround to this issue, as does using only short timeouts on queries.